LMDS, MMDS race for low-cost implementation
By Loring Wirbel
EE Times
(11/30/99, 3:02 p.m. EDT)
Advocates for wireless residential broadband access had a busy November, as the millimeter-wave camp examining 28- and 38-GHz local multipoint distribution service (LMDS) networks raced to make consumer and business systems cost-effective for single-user Internet access. They faced an onslaught of new efforts from competing radio system specialists aiming to make the microwave networks known as multichannel multipoint distribution service (MMDS) at 2.5 GHz and 5 GHz easier to deploy. Cost-reduction efforts in both markets may result in making wireless "last-mile" access technologies viable alternatives to always-on Internet access technologies such as cable modems and digital subscriber lines.
The IEEE's new 802.16 working group for broadband wireless access networks is trying to hammer out common principles for both MMDS and LMDS systems. At a meeting in Koloa, Hawaii, in mid-November, the panel worked on physical-transceiver, Medium Access Control and frequency-coexistence standards for both camps. On Nov. 11, the working group approved an official study group extension to examine MMDS and similar networks that operate at less than 10 GHz. The existing 802.16.1 air interface project is for networks ranging in frequency from 10 GHz to 66 GHz.
Earlier in the month, Cisco Systems Inc. (San Jose, Calif.) unveiled a vector orthogonal frequency division multiplexing coalition for lowering MMDS client costs, claiming that the use of VOFDM access would make MMDS systems more reliable than previous 2.5-GHz systems (see Nov. 1, page 1). This could place the non-line-of-sight microwave technology well ahead of LMDS, which requires a line-of-sight antenna network resembling cellular grids in its infrastructure. Cisco has lined up Broadcom Corp. (Irvine, Calif.) and Texas Instruments Inc. (Dallas) to make access controllers for VOFDM, borrowing heavily from cable-modem chip sets in order to make wireless broadband chips as low-cost as possible.
OFDM modulation is no stranger to low-cost networking. Wi-LAN Inc. (Calgary, Alberta), for example, introduced a wireless local loop technology last summer using its own OFDM derivative, Wideband OFDM. The "iWiLL" system can be used in bands such as 2.5 GHz and 3.5 GHz, modified to provide voice local-loop and broadband Internet access from a common platform.
The next entrant to the field will be showing up in early 2000, when Malibu Networks Inc. (Calabasas, Calif.) rolls out an OFDM-based MMDS system, developed independently from Cisco's VOFDM coalition effort. Bill Baker, chief executive of Malibu and the founder of router and DSL specialist Sourcecom Inc., was at the Next Generation Networks conference in Arlington, Va., in mid-November, quietly telling investors about Malibu's plans for wireless broadband.
Patents galore
"LMDS will be great for five years down the pike, so we really don't want to position our company as anti-LMDS," Baker said. "What we find, however, is that it's much more interesting to look at the 2.5-GHz band, and the FCC's 5-GHz U-NII band, if you want to remove regulatory overhead and tap into the availability of lower-cost RF components."
Malibu has received 21 patents and filed 67 patent claims, that center on proprietary concepts for optimizing wireless channels for Internet Protocol (IP) transport. The company sees a lot of value in standard Medium Access Controllers (MAC) and standard modulation schemes, and Baker said Malibu largely supports the Cisco concept for combining OFDM with MAC chips that follow the Docsis model. Malibu will make its proprietary mark in what it calls WINAAR (Wireless IP Network Access Architecture) IP prioritization technology.
Other companies founded to optimize wireless local loop systems for voice applications are developing new spin-off architectures for broadband data access in MMDS bands. Spike Technologies Inc. (Nashua, N.H.), for example, is designing special versions of its Prizm broadband delivery system for data applications in the 2.1-GHz to 2.7-GHz bands, as well as the 3.4-GHz to 3.7-GHz bands. The company has longer-range plans to use its sectored antenna designs and V2 radio in LMDS systems, but is finding more near-term interest in the lower bands.
The 2.5-GHz industrial, scientific, and medical band also has been an arena of previous data activity. Most hardware developers, however, have focused on using 2.5-GHz RF chip sets for in-building wireless LANs, using either a single radio access point, or a couple of simple bridges for multibuilding LAN access. The one exception to the rule outside MMDS realms has been Metricom Corp. (Los Gatos, Calif.), which has used a metropolitan network of 2.5-GHz access points to create a public-network 128-kbits/second Internet access system.
Metricom's Ricochet service has been offered for several years, with moderate take rates in the Bay Area. It has received some important leases on life in recent months, however. Microsoft founder Paul Allen invested in Metricom in late summer, and Sierra Wireless Inc. (Burnaby, B.C.) signed a licensing pact with Metricom Nov. 9, in which the Canadian company would provide PC Card modems for the Ricochet service. MMDS advocates, however, call a 2.5-GHz, 128-kbit meshed network service limited to only handheld platforms, since its speed does not reach the megabit speeds of VOFDM-based services.
Minimum impact
Is all this having much of an impact on LMDS rollout plans? Not much, according to Carlton O'Neal, vice president of marketing at LMDS startup Ensemble Communications Inc.
(San Diego). As a licensed service having to deal with the hassles of line-of-sight antenna placement such as roof access to public buildings, LMDS technology was always anticipated to take longer to deploy than MMDS, O'Neal said. But its ability to scale to speeds common in the DSL and cable-modem worlds makes the service much more attractive than those in lower frequency bands, according to Ensemble executives.
Ensemble and its Bellevue, Wash.-based competitor Wavtrace Inc. represent the new breed of LMDS specialists who seek to apply second-generation packet-centric services, such as time-division duplexing and adaptive algorithms optimized for Internet Protocol, to the problem of building effective local access markets.
Ensemble's Fiberless Broadband systems, on display as full prototypes for the first time at last month's Telecom '99 show in Geneva, offer multiple layers of adaptive treatment of IP packets in order to allow access points to dynamically sculpt bandwidth according to the needs of users. This feature set is important for implementing IP quality-of-service parameters in a wireless environment.
According to Cisco's VOFDM advocates, it also may be necessary to handle what the VOFDM coalition sees as an inherent disadvantage of LMDS. Slow synchronization to rapidly varying radio signals can be a problem for LMDS, said Steve Smith, director of broadband wireless networks at Cisco, because consumers expect broadband Internet access to be an always-on service. They may leave their computers for hours at a time, then come back to their LMDS-enabled computer expecting instant access. If MMDS using VOFDM shows faster synchronization in such environments, as Cisco anticipates, it may chalk up another advantage over its higher-frequency cousin.
In any event, developers are anxious to get common MAC-layer silicon approved that can be used in both subsets of the 802.16 standard. Roger Marks of the National Institute of Standards and Technology's N-WEST program, who helped found 802.16, said that the IEEE working group is looking to a model of proposing a variety of physical-layer standards, linked to common air interfaces for lower and higher frequencies. |
